\textbf{A. Regression discontinuity} & (1) & (2) & (3) & (4) & (5) & (6) \\ \midrule
\hspace{0.1cm} Born after reform&     0.08\sym{***}&     0.05\sym{***}&     0.04\sym{***}&     0.03\sym{**} &     0.01         &     0.00         \\
                &   (0.03)         &   (0.02)         &   (0.01)         &   (0.01)         &   (0.01)         &   (0.01)         \\
\midrule
Bandwidth       &        2         &        5         &       10         &       15         &       20         &       30         \\
Outcome mean    &     0.20         &     0.21         &     0.23         &     0.25         &     0.28         &     0.35         \\
Observations    &    10827         &    22689         &    44124         &    72740         &   106205         &   190696         \\
\midrule \textbf{B. Household fixed effects} & (1) & (2) & (3) & (4) & (5) & (6) \\ \midrule
\hspace{0.1cm} \textit{Reform}&     0.04\sym{**} &     0.04\sym{***}&     0.04\sym{***}&     0.06\sym{***}&     0.07\sym{***}&     0.10\sym{***}\\
                &   (0.02)         &   (0.01)         &   (0.01)         &   (0.02)         &   (0.02)         &   (0.03)         \\
\midrule
Bandwidth       &        2         &        5         &       10         &       15         &       20         &       30         \\
Outcome mean    &     0.12         &     0.12         &     0.12         &     0.13         &     0.14         &     0.19         \\
Observations    &     5419         &    25417         &    78027         &   155313         &   252715         &   543094         \\
